UMass Calibration System User Guide

Table of Contents

• Description of Programs
  
  
• Using the Calibration System
  
  
• Using the Dot Finder
  
  
• Using Calibrate
  
  
• Using Undistort
  
  

Description of Programs

There are essentially two different programs that make up the system. The first is the calibrationsystem program which uses calibration images and camera parameters to generate a calibration report. The second is the Undistort program which can undistort a TIFF image given a calibration report for the camera that produced it. There are also two variations of this program, Undistort4band and Undistortpgm which work on 4-band TIFFs and 1-band PGM files, respectively.

Using the Calibration System

There are two stages to the calibration process: finding dots in the calibration images and performing the actual calibration. Both of these steps are accomplished by the calibrationsystem executable. To begin, run calibrationsystem on Linux, or calibrationsystem.exe in Windows.

Using the Dot Finder

1. Click the Browse button next to the Image File box to select the first, closest calibration image. Once open, the image will be displayed in the window.
   
   
   Step 1. (Click Image to Enlarge)
   
   
2. In the area where the calibration image is displayed, click with the left mouse button and drag to draw a blue box around a "typical" dot. This dot will be used to find the other dots in the image; it is better to draw a box that is somewhat larger than the dot rather than trying to get a tight fit. Release the mouse button when you are satisfied with the size of the box. Once this step is complete, the Window Size parameters will be automatically updated to reflect the size of the box. These parameters should not be manually edited.
   
   
   Step 2. (Click Image to Enlarge)
   
   
3. Again in the calibration image display, click with the right mouse button and drag to draw a red box around the area of the image where all of the dots appear. This defines the area where the dot finder will search for dots, so it should generally include the entire calibration target area and nothing else. Rows that include dots which are clipped by or lie very close to the edge of the image should not be included. Defining the area in this way will yield a faster, more accurate search. Once the button is released, values will be placed in the two pairs of Row and Col boxes. These values should not be manually edited.
   
   
   Step 3. (Click Image to Enlarge)
   
   
4. In the Limit box, enter a threshold to use to differentiate between pixels belonging to dots and those which do not. Typically a value of 220-230 works well. A lower value will increase the number of dots that are found, but could also introduce false positives.
   
   
   Step 4. (Click Image to Enlarge)
   
   
5. Enter the correct value in the Dots in a Row box. For the UMass calibration target, this number is 9. The correct value will be automatically entered into the Cols box.
   
   
   Step 5. (Click Image to Enlarge)
   
   
6. Specify the desired location for the XYZ output file box by clicking Browse and locating the directory where you would like the calibration data to be stored. By default, the data is saved to the directory where the image is located.
   
   
   Step 6. (Click Image to Enlarge)
   
   
7. Enter the number of visible rows of dots in the current image in the Rows box. Important: this number should correspond to the number of rows that you drew a red box around in step 3. So if you did not select a row of dots because it was clipped by the edge of the image, do not include this row in the count.
   
   
   Step 7. (Click Image to Enlarge)
   
   
8. Enter the appropriate numbers in the Startrow (in dots) and Startcol (in dots) boxes. The lower left visible dot on the calibration target is at location (0,0). You should enter the coordinates in dot rows and columns of the lower left visible dot in the image. For example, if the lowest row of dots on the calibration target is not visible in the image, but the full width of each row can be seen, the Startrow should be 1 and the Startcol should be 0. Important: again, ignore any rows that were not selected in step 3.
   
   
   Step 8. (Click Image to Enlarge)
   
   
9. Enter the distance in meters between dot centers on the calibration targetin the Step box. For the UMass calibration target, this value is 0.192m.
   
   
   Step 9. (Click Image to Enlarge)
   
   
10. Enter the distance in meters from the closest setting of the target on the calibration table in the Z box. Typically, this is 0m for the first image, 0.1m for the second, and so on.
    
    
    Step 10. (Click Image to Enlarge)
    
    
11. Click the Run Dot Finder and MakeXYZ button. After a few moments, the display will show the locations of the dot centers as determined by the dot finder. You should them examine the results of the dot finder; the white cross hairs should be in the centers of every dot on the target. If not, you may need to adjust the boxes you drew. To do this, click Go Back and rerun Dot Finder and repeat steps 2 and 3, and then return to step 11.
    
    
    Step 11. (Click Image to Enlarge)
    
    
12. If you are satisfied with the dot finder's results, click the Append Files button.
    
    
    Step 12. (Click Image to Enlarge)
    
    
13. Repeat steps 1 to 12 for each of the calibration images. Once all images have been processed, click Go to stage 2 (Calibrate). Make sure you click Append Files on the last image before going to stage 2.
    
    
    Step 13. (Click Image to Enlarge)
    
    

Using Calibrate

1. Click the Browse button next to the XYZ File box and locate the XYZ file created during the first stage.
   
   
2. Click the Browse button next to the Row Col File box and locate the Row Col file created during the first stage.
   
   
3. If you have a parameter file for your camera, click Browse next to the Parameter File box and locate the file. If you don't have a parameter file, click the Create button, specify a name and location for a new parameter file, and enter the appropriate camera parameters as supplied by the manufacturer in the window that appears.
   
   
4. Click the Run button. The calibration results will appear on the screen and will be stored in a file called parout.txt in the current directory. You can now exit the program.

Using Undistort

1. Use the New command from the File menu to generate a list of images from the contents of a directory, or use the Open... command to load a text file that you have previously created containing a list of images to process.
   
   
2. Use the Browse button to locate the parameter file created in stage 1.
   
   
3. Click Run. The undistorted images will be placed in the same directory as the original images, with _und appended to the file name.
This Page Copyright © 2005 Computer Vision Laboratory, University of Massachusetts-Amherst